Temperatwe Dehydration Studies of Octacalcium Phosphate
1631
Programmed 'Temperat ure De hy dra t ion St udies o f 0ct a calc ium Phosphate C.William Anderson, Ralph A. Beebe, and J. S. Kittelberger*' Department of Chemisfry, Amherst College, Amherst, Massachusetts 01002 (Received January 4, 7974) Publication costs assisted by the National lnsfitutes of Health
Thermal dehydration spectra of octacalcium phosphate (OCP) are obtained with a mass spectrometric technique a t linear heating rates. Infrared spectra of vacuum-pyrolyzed OCP are obtained, as well as BET surface areas and pyrolysis weight-loss measurements in v'acuo. Infrared results indicate a growth of OH - and HP0i2- concentrations at temperatures above -200", growth of P ~ O Tconcentration ~at higher temperatures reaching a maximum near 500" and falling off near 700°, and a concommitant loss of OH - and P 2 0 ~ ~ in -the 600-800" region. Six discrete processes are identified in the thermal dehydration spectra. One of these, occurring near 200", is interpreted as the disappearance of the OCP lattice, accompanied by a large loss of water and growth of OH- and HP042- concentrations. The activation energy for this process is measured to be 59 f 5 kcal/mol. A second, occurring about 700", accompanied by simultaneous loss of OH- and P2074-, is probably the reaction Calo(P0c)dOH)z + C a ~ P z 0 7 4Can(P04)2 HzO. The activation energy for this process is found to be 122 f 11kcal/mol.
-
+
Introduction This is one in a series of studies of the state of water associated with bone mineral and various synthetic calcium phosphates. In previous work, Dry and Beebe measured heats of adsorption of water on anorganic bone and a synthetic hydroxyapatite (HA), and found values ranging from 22 to 11 kcal/mol.2a Holmes and Beebe found evidence for considerable internal water in synthetic amorphous calcium phosphate (ACP) but for only surface water in HASzbSedlak and Beebe3 recently studied the internal water in ACP by an extension of the programmed temperature desorption (TPD) method developed by CvetanoviE and Amenomiya.4 Sedlak has established the validity of the TPD technique for obtaining approximate activation energies for bulk-dehydration processes in solids.5 Prior to the work in this laboratory the use of TPD had been confined to surface desorption phenomena. The intention in the present work was to apply these methods to octacalcium phosphate (OCP) in order to learn more about the decomposition processes and binding energy of water in this highly hydrated material. Fowler, Moreno, and Brown6 have reviewed the literature on OCP, which indicates the occurrence of this substance in association with other calcium phosphates both in vitro and in vivo These authors have studied the pyrolysis of OCP using ir spectroscopy, weight loss, and chemical analysis to characterize the pyrolysis products up to 900". Weight loss up to 700" approximates that expected from the formula for OCP: ca&(PO4)66H20, on the assumption that water i s the sole gaseous pyrolysis product. In the presient study we follow the dehydration of OCP by the mass spcctrometric thermal analysis (MTA) technique which is closely related to TPD or to the original flash desorption m e t h c ~ dIn . ~ MTA the sample is heated in UQCUO, so that (in our apparatus) the temperature rises at a linear rate. G,aseous decomposition products are monitored with a mass spectrometer, most of the interest in this work being in WzQ. It has been that in a first-order desorption, under conditions obtained in our apparatus, the shift in the temperature of a peak (TM) in the MTA spectrum, with heating rate @, obeys the equation
In (TM2/@)= EA/RTM f In ( E A / A R ) where EA is the activation energy for the desorption, A is the Arrhenius frequency factor, and 13 is the gas constant. Thus, if the results of a series of experiments at different heating rates are plotted in the form In ( T d / @ us. l / T M , a straight line is obtained with slope EAIR. Recently Lord and Kittelberger have shown8 that the same graphical method can be used to obtain activation energies in second (and probably higher) order processes. For fixed @,the position of a peak (TM)is sensitive to the initial concentration of reagent in higher order, but not in first-order reactions; this phenomenon can be used to determine reaction order.s Wightman has shown that in a situation where a distribution of activation energies exists, a rather broad, flat-topped desorption peak occurs.1o The present work, therefore, extends the earlier work of Fowler, Moreno, and Brown,6 and allows the measurement of the activation energies of the discrete dehydration processes.
Experimental Section The MTA runs were carried out in a two-chamber highvacuum system shown in Figure 1. The sample chamber is a 36-1. aluminum cube with Viton-A O-ring seals. This chamber is pumped by a 7-in. oil diffusion pump (CVC) baffled by a liquid nitrogen cooled trap (Granville-Phillips), and the estimated pumping speed at the throat of the sample chamber is 400 l./sec in the Torr range. The sample chamber walls are partially shrouded by an additional liquid nitrogen trap to reduce background pressure. Base pressure in this chamber is 1 x Torr from a Bayard-Alpert gauge, and rises to a typical maximum of 5 x lo-? Torr during a run. The sample chamber is separated from the detector chamber by a 2-in. bore "butterfly" valve (Edwards High Vacuum). This chamber is a 3-1. copper-brazed stainless steel cylinder sealed with Conflat flanges and QFHC copper gaskets (Varian), and pumped by an 8 l./sec ion pump (Varian). This chamber houses a quadrupole mass spectrometer (EA1 Model 250) equipped with a copperberyllium particle multiplier. With all settings optimized, The Journalof Physical Chemistry, Voi. 78. No. 16. 1974
C. W . Anderson, R . A. Beebe, and J. S. Kittelberger
1632 I
n
/
/'
,I
1s I3
apparatus: 1, ionization gauge: 2, monitor thermocouple: 3, oven: A , sample compartment: 5, butterfly valve: 6, ionizer assembly: 7, quadrupole filter: 8, mass spectrometer control panel: 9, thermocouple gauge; 10, vacuum chamber: 11, sensing and power cables; 12, ion p u m p ; 13, temperature programmer: '14, deiector/multiplier;15, two-channel recorder. Figure 1. MTA
this instrument is capable of detecting a partial pressure of Ng of Torr, The sample oven ir; a cylinder 3.7 cm in diameter by 6.3 cm long made of OFHC copper with a 1.0-cm diameter sample cavity bored 3.0 cm deep on the oven axis. The heater consists of 12 spirals of 0.3%" tantalum wire in series, mounted in longitudinal holes around the sample cavity, and insulated by thin-walled alumina tubes. Operating at maximum temperature (-1800°K) and 120 V ac, this heater dissipates more than 300 W. The oven is positioned within a stain1,ess steel radiation shield by four thin stainless screw3, the whole oven assembly is mounted on an adjustable stainless arm so that the sample cavity is coaxial with the mas3 spectrometer ionizer axis. Temperature measurement and control are achieved with two chromel-alumel thermocouples screwed into the copper oven, which agreed, after calibration, to within the precision of the measuring equipment, f0.2". In order to determine the magnitude of any thermal lag which may exist between the copper oven and the powdered sample, about 10 mg of which is spread thinly directly on the oven floor in a typical run, we have compared dehydration peak temperatures obtained in this apparatus with those obtained in a TPPr apparatusll for the same sample and beating rates. In 'TPX), the sample is heated in a stream of helium which :s preheated in a baffle at the front of the oven; no thrmnal lag exists with this arrangement ,I1 Peak temperatures obtiained by the two methods at identical heating rates agree within fl" at the highest heating rates, ensuring, we believe, that any thermal lag which may exist between oven and sample in our MTA apparatus is no greater than the Beast uncertainty in locating the peak maxima, which is shown below to be fl". Ten heating rates between 0.5 and 30"/min are provided by a Hewlett-Paekard Model 240 temperature programmer. In thic, unit a voltage derived from a motor-driven potentiometer is opposed to the output from one of the thermocoup1.es. the amplified error is used to gate an SCR which powers the oven heaters. The limiting source of error in sample heating rate appears to be the nonlinearity in the emf us. temperature curve for chromel-alumel. This has a very slight S-shape in the temperature region studied, with a variation in slope of f 3 % , which is the vaxiation we find in sample heating rate during the course of a run. The actual heating rate is obtained for each peak as the tangent to the thermocouple emf us. time plot at the exact time of the peak. In order to facilitate this determination of heating rate, The Journalof F'hyiical Che!rnistry, Vol. 78, No. 16, 1974
E-
o
100
pi
200
300
500
400
600
700
3~ I m n
800
900
T, 'C
Figure 2. MTA spectra of Ha0 lost from OCP.
the mass spectrometer signal and the signal from the second thermocouple are simultaneously plotted on a twopen recorder. During each run the temperature pen is calibrated frequently against a Leeds & Northrup potentiometer. The overall uncertainty in determining the temperature of a peak depends on a number of factors, sharpness of peak, spectrum noise, recorder scale, etc., but for no peak reported here does this exceed f2", and it drops to fl"in favorable cases. The OCP sample studied here was prepared by the slow hydrolysis of CaHP04.2H20 in 1 M sodium acetate at 38".12 Chemical analysis of the sample gives 30.7 wt % Ca (EDTA titrationls), 55.5 wt % PO43- (spectrophotometric phosphomolybdate method14), or a Ca/P molar ratio of 1.37 0.04; the expected ratio for stoichiometric OCP is 1.33. No carbonate analysis was performed, but the ir spectrum shows no detectable carbonate or bicarbonate bands. Infrared spectra were obtained on samples dispersed to about 5 wt % in dried KBr, and pressed in a small hand press. Where H2O is of special importance, an identical pellet of pure KBr can be used in the reference beam. Spectra were recorded on a Perkin-Elmer Model 237. The ir spectrum of the unheated OCP samples i s shown below in Figure 3, and is identical with spectra reported previously by Fowler, Moreno, and Brown.6 Our sample shows shoulders at 3570 and 633 c m - I , suggesting the presence of traces of HA. X-Ray powder diffraction patterns o l the sample also confirm its identity as OCP. In particular, the 18.7-a reflection15 most characteristic of OCP is very strong in the X-ray patterns of the OCP studied here. To complement the MTA spectra obtained here, the surface area of the OCP has been determined by BET nitrogen adsorption.l6 The samples were outgassed at Torr overnight at the temperatures indicated in Table I, which summarizes the results.
*
Results a. MTA Studies. The dehydration spectrum of OCP obtained with the mass spectrometric apparatus is shown in Figure 2. We have reproduced here runs at heating rates of 3, 10, and 30"/min in order to show how the peaks shift with heating rate. Background spectra of the empty oven are featureless except for a burst of water from the heaters and another from the alumina insulators which occur immediately after starting the run (at temperatures below loo"); these have been suppressed in Figure 2. The peaks
Temperature Dehydration Studies of Octacalcium Phosphate
1633
T, L M
-
L
uoo
-
-
L
HDol
P5w
1
2000 ~ J o 0 l y z ) 1 w o l l 0 0 loo0 ')o 800 WIVENUMBER, C M '
Figure 3. Ivlfrared spectra of OCP heated to tures in vacuo: A, unheated; B, 157"; C, 271": The absorbance scale is logarithmic.
mo
600
5w
various temperaD, 625"; E, 800".
TABLE I: Effect, off Degassing Temperature on Surface Area of OOP as Measured by BET Method ~~~
Outgassing temp, O C
Surface area, ml/g
_ I _
22
100 200
300 500
4.7 3.6 5.2 5.4,5.7 7.0, 6 . 5
are numbered in order of appearance from low to high temperature a8 X through VI. The spectra in Figure 2 have been normalized to the height of peak 11, not to unit sample weight. The spectra, thus, do not allow observation of changes in the area of peak I1 with P. We find that the area under peak I decreases somewhat with increasing evacuation time before the start of the run, and the position of the maximum is also sensitive to these prepumpxiig conditions. Peaks I11 and IV are least pronounced at low p, and increase in area as @ increases. Peak VI appears with smaller area a t the greatest heating rates. No quantitative measure of the areas under the overlapped peaks was attempted; while curve shapes for simple, single-step desorption processes are understo0d,8,~there is no a priori reason to suppose that the processes under investigation are simple arid single step. b. Weight Loss. Samples of the OCP studied here lost 10 f 1% in weight during the course of a run to 850", essentially all of which was HzO, since the mass spectrometer detected negligible amounts of other gaseous effluents (such as 602) during a run. An unknown fraction of the weight loss can easily occur during pumpdown before the start of heating. Thus, the spectra in Figure 2 cannot be assumed to record all of the water lost from the sample; very loosely-bound water will not be recorded by this high-vacuum technique. For reference, however, a stoichiometric OCI' contains 9.16% water of hydration, with additional loss possible uia pyrophosphate formation and desorption of suitface Rater. e. Infrared Studies. To assist the interpretation of the dehydration spectra, we have recorded, at room temperature, infrared spectra of samples heated under high vacuum to various temperatures in the sample oven of the MTA apparatus. All of the spectra shown in Figure 3 are of samples held a t the temperature indicated for 7-8 hr
*C
Figure 4. Infrared band intensities of OCP samples heated to various temperatures: H 2 0 , 0; H P 0 4 * - , a; OH-, 0 ; P 2 0 7 4 - ,
e.
and heated to this temperature at 15"/min. These spectra are in essential agreement with those reported by Fowler, Moreno, and Brown.6 We have monitored the concentrations of four species during the course of these pyrolyses: XZO,OH-, P z O ~ ~ - , and HP0d2-. To do so we have measured the intensities of the ir bands at 1630 cm-1 for HzO, at 3570 cm-l for OH-, at 725 cm-I for P20T4-, and at 870 cm-1 for HPO&. The band assignments have been disc ~ s s e d . ~ J 7 J 8The reported band intensities were measured with a planimeter and normalized either to unit sample weight or to unit area under the phosphate ir band between 500 and 700 cm-I. The two methods of normalization agree within the reliability expected for this kind of procedure. Before reporting our findings here, we caution the reader that this method of determining ion concentrations in a pyrolysis experiment is beset with several difficulties, and cannot be regarded as completely quantitative. For example, extinction coefficients vary with environment, a well-known problem for the 3570-cm-l OHline.lg The qualitative trends indicated by the ir intensities in Figure 4, however, are certainly meaningful. Other bands could have been used to monitor the ion concentrations. As a check, we have measured the intensities of the 633-cm-l OH- libration. For all spectra, OD(3570)/ OD(633) = 0.26 rt 0.06. The errors, we feel, can be due entirely to the area measurement. This figure shows a simultaneous increase in both I I p O d 2 - and OH- in OCP samples heated to 157", and to 271". OH- then remains reasonably constant to 625" and falls off in samples heated above 625". HP042- falls off above 271" in the region where p ~ O 7 ~rises. - HzO falls off steadily, and is nearly absent from samples held for 8 hr at 450" or higher. Since the heating time in all the above ir pyrolysis studies (-8 hr) was considerably longer than the duration of a typical MTA run (-1 hr a t @ = 15"/min), we have conducted two ir experiments in which the total heating time was approximately 1hr. In the first a sample was heated at 15"/min to 359" and cooled immediately to room temperature. The ir spectrum of this sample showed significant increases in OH- and HP042- over the amount present in the original material and less of these ions than found in samples heated to 271" for 8 hr. In this same experiment no p ~ 0 - z ~ was found. In the second experiment a sample was heated to 800" The Journaiof Physical Chemistry. Vol. 78, No. 16, 7974
C. W. Anderson, R. A. Beebe, and J. S. Kittelberger
1634
11.500
l1.000
I I
+10.500
0.972 1000/Tu,
Figure 5. Ln(TM:'/@) vs. l O O O / T h , for
I i in
the MTA spec-
TABLE I1 _ I _ _ -
Peak II Tu,' C
3.3 5.4 10.4 10.5 13.8 16.2 18.3 22.1 24.6 31.6
218 220 228 228
Peak V I Tar,
O C
721 743 742 746
230 747 231 752 236
at 15"/min arid cooled immediately. The resulting ir showed more of both OH- and P 2 0 ~ ~ than - in the sample heated for 8 hr a t $OW, but significantly less than in the sample heated at 62;5" for 8 hr. The loss of OH- and P20T4- between 600 and SOW would thus appear to be coupled. In comparing our results with those found in other pyrolysis studiles, the ri3ader should bear in mind that our samples were heated i,n vacuo at loe7 Torr. d. Actiuatior; Energies. Attempts were made to determine E A for all six peaks appearing in the MTA spectrum of OCP by plotting In (TbI2/O)us. 1/TM. Only for peaks I1 and VI are t.hese plots satisfactory; for the others there is such a large standard deviation in the slope of the line that we feel the results are meaningless. The data for peaks I1 and VI are recorded in Table 11. The plots for peaks 11and VI are pimented in Figures 5 and 6. The slope of the least-squares line gives an activation energy of 59 kcal/mol with a standard deviation of 5 kcal for peak 11. For peak VI E:%:= 1.22 kcal/mol with a standard deviation of I1 kcal. iscussion The infrared studies show that at fairly mild temperatures (150-350") the concentrations of HP042- and OH- increase This suggests that the simultaneous increases in both may arise from a single chemical reaction
5,O-+ P04J- = HPO,*-
+ OH-
in which some of the water of hydration is consumed. This could arise from the formation of separate phases of CaIIPOc and hydroxyapatite -+
0.996 OK
peak
p, deg/min
0.988
~oooiru,
trum of O W .
C a A (PO,), 5H 0
0.980
'K-'
+
XCa,,(P04)&OH), 3CaHP0,
+ 4H.0
GaHP04 has been found as a pyrolysis product of OCP in The Journal of I'hy!jicai Chsmistry, Vol. 78. No. 76, 1974
Figure 6. L n ( T M 2 / @ )
'.004
1.012
I
vs. 1000/TM for peak V I in the MTA spec-
trum of OCP.
this temperature region in some studies,I2 but not in 0thers.6 This may be due to different experimental conditions such as partial pressure of water and duration of pyrolysis. Our results indicate no discrete CaHP04 phase; the doublet a t 1350-1400 cm characteristic of CaHP046 never appears in our ir spectra. It is worth noting that those studies which found CaHPQ4 used X-ray diffraction,12 while no conclusive ir evidence has been found for CaHP04 in a pyrolysis study.12 Another possibility, favored by some authors,20 is the formation of a defect, or calcium-deficient apatite; for example CasH,(PO,), . 5H20
--+
Ca8H,(P0,)sOH . 4H20
Since the MTA technique is a nonequilibrium technique, and since we have no X-ray patterns of pyrolyzed samples, we cannot draw much information from our data on the question of what solid products are formed in this temperature range. With the infrared results in mind, we propose the following assignment of the MTA spectrum. Peak I is the first step in the loss of water of hydration. Given the sizable area under this peak together with the fact that some loosely held water is lost during pumpdown, we can state that at least 20% of the total dehydration is associated with this step. Fowler, et al., found roughly 50% of the weight loss occurred below 150°.6 The tendency of TMto shift to higher temperatures with greater prepumping time is due, we conclude, to the fact that a distribution of values of EA exists for this process. Loss of water with low EA during pumpdown would shift TMto a higher value. Peak 11, the largest in the spectrum, occurs near 22Q"; mod2-and OH- are being formed in the same temperature range. OCP heated to 200" begins to lose its structural identiiy, and HA and CaHP04 have been reported to appear near this temperature.12 We assign this peak as the loss of the bulk of the water of hydration in OCP. The activation energy associated with this process, 59 kcal, is considerably greater than the 20-30 kcal activation energies found for desorption of water from HA surfaces,21 so that we believe peak I1 cannot be due to a desorption process. Peaks 111 and IV grow in area as the beating rate is increased. This fact suggests that the water lost here can also be removed by a competitive chemical reaction; at low heating rates more of the water is removed by the competitive mechanism, while at high heating rates more
Temperature Dehydration Studies of Octacalcium Phosphate
of it is lost in peaks I11 and IV. We note that peaks V and VI decrease in area as /3 increases suggesting that reactions V and VI are the terminal steps in the alternate mechanism. Our proposed assignments for peaks 111 and IV are two additional stages in the loss of water of hydration. The grcater the heating rate the more undecomposed OCP remains a%these elevated temperatures. Peak V we assign as the water loss associated with the formation of pyrophosphate
--
2HPQd2-
+
P2074- H20
which is known to occur in this temperature region.6,22 Fowler, et al , for example, find a maximum yield of pyrophosphate in OGP samples pyrolyzed at 550°,6 very nearly the temperature of peak V a t low p. Peak VI occurs in a temperature region where the present infrared study and earlier work6 indicate losses of P2074- and OH-. Fowler, et a1.,6have suggested that the reaction Ca,o\P0,),(OH)2
+ CazP207
4Ca,(PO,),
--+
+ H20
accounts for weight loss near 700". In the present study we find a simultaneous) decrease in P ~ 0 7 ~ and - OH- in samples heated to 800"; if peak 111 were due to simple dehydroxylation no decrease in P ~ 0 7 ~ would be expected. In addition the ir spectrum of samples heated to 800" shows bands at 1120, 972, 950, 602, 589, 550, and 541 cm-1 which have been reported in @-Ca3(P04)2.6We find an activation energy of 122 kcal for this reaction which i s known to be very S ~ ~ O W Our . ~ assignment of peaks 111-VI is consistent with the observation that peaks I11 and IV increase in area with increasing 6, while peaks V and VI decrease with ~ncraacing @. Quantitative measure of the areas of these peaks is net possible, but it is worth noting thac the weight loss3 during a complete MTA run seems independent o f @ . Our weight ~ O S Sstudies give a 10% decrease in weight on heating to 850", which corresponds to about 5.5 mol of water per mole of OCP. This finding is in essential agreement with data reported by Fowler, et al.6 From Table I, the specific surface area of the OCP is on the order of 5 m2/g. Using 11 A2 as the cross sectional area of the water molecule,2 a close-packed monolayer of adsorbed water on the OCP would represent less than 0.01 mol of HzO per mole of OCP. Adsorbed water, thus, represents a negligible fraction of the 10% weight loss. We suggest that this dehydration can be understood by the following set of reactions, consistent with the above assignment of peaks
for a total loss of 6 mol of water. Since 1 + X/2 + X/2 or 1 + X of the 6 mol of water are lost in the last, slow iteps, we can readily see how only the 4 mol of water of hydration plus approximately 1 of the 2 mol of water available for
1635
loss in the last two (slow) steps would show up in our weight loss measurements, in which the total heating time was 56 hr. Fowler, et al., also observed 10%weight loss6 in pyrolyses of much longer duration but heated their samples in a muffle furnace at atmospheric pressure. Not much is said in this discussion about the crystalline phases formed in the decomposition of OCP. Our experiments were not designed to produce maximum yields of the decomposition products, and we have limited structural information. There is evidence for HA in the parent sample, and increasing amounts of it above 200". There is evidence for @-Ca3(P04)2 at the highest temperatures reached in our experiments. Evidence on the formation of discrete phases of CaHP04 and CszzP207 is less clear in the present study. Note Added in Proof. In preparing the manuscript, we overlooked a relevant reference on this subject.23 Newesely cites evidence for one of the five water molecules being held rather loosely in the OCP lattice, so that its loss may be reversible. The loss of such a loosely held water molecule would either be complete during the prepumping stages of our MTA experiments, or it could account for peak I in the MTA spectrum. Additionally, Newesely finds P2074formation has a threshold temperature near 150" in pyrolyzed OCP. This is in essentially perfect agreement with our findings (see Figure 4).
Acknowledgments. We wish to thank J. M. Holmes for carrying out the surface area analyses, and A. S. Posner for providing the OCP sample used here. We thank Aaron S. Posner and John M. Sedlak for helpful discussions. Financial support from NIH Grant No. AM-2896 is gratefully acknowledged. References and Notes (1) Address correspondence to Xerox Corp., Hldg. 114, 800 Phillips Road, Webster, N. Y. 14580. (2) (a) M. E. Dry and R. A. Beebe, J. Phys. Chem., 64, 1300 (1960): (b) J. M. Holmes and R. A. Beebe, Calcif, Tissue Res., ?, '163 (1971). (3) J. M. Sedlak and R. A. Beebe, J. Colloid lnterface Sci., submitted for publication. (4) R. J. CvetanoviC and Y. Amenomiya, S. Phys, Chem., 67, 144 (1963), (5) J. M. Sediak and R . A. Beebe, Thermochim. Acta, submitted for Dublication. (6) B 0 Fowler, E C Moreno, and W E Brown, Arch Oral B ~ o l 11, , 477 (1966) (7) G. Ehrlich,'Advan. Catal., 14, 256 (1963). (8) F. M. Lord and J. S. Kittelberger, Surface Sci., 43, 173 (1974). (9) P. A. Redhead, Vacuum, 12, 203 (1962). (IO) Wightman's results are unpublished but cited extensively by R . J. CvetanoviC and Y. Amenomiya, Catal. Rev,, 6, 21 (1972). (11) J. M. Sedlak, unpublished results. The spectrometer and temperature measurement in it is described in ref 3. (12) W. E. Brown, J. R. Lehr, J. P. Smith, and A. W. Frazier, J. Amer. Chem. SOC., 79, 5318 (1957). The sampie used here was supplied by Dr. A. S. Posner, Cornell Medical College. (13) R. G. Yalman, W. Bruegemann, P. T. Baker, and S. M. Gam, Anal. Chem., 31, 1230 (1959). (14) F. J. Welcher, Ed., "Standard Methods of Chemical Analysis," Vol. lib, 6th ad, Van Nostrand, New York, N. Y., 1963, p 2467. (15) W. E. Brown, Nature (London), 196, 4859 (1962). X-Ray patterns were run by Dr. A. S. Posner. (16) Surface areas were determined by Professor J. M. Holmes, Carleton University, Ottawa, Canada. For a description of the BET technique as applied to calcium phosphate systems, see ref 2b. (17) See S. H. Joris and C. H. Amberg, J. Phys. Chem., 75, 3167, 3172 (1971), for a d,iscussion of the assignment of this band. (18) W. E. Klee and G. Engel, J. lnorg, Nucl. Chem., 32, 1837 (1970) (19) N. C. Blumenthal and A. S, Posner, Calcif. Tissue Res., in press. (20) E. E. Berry andS. A. Leach,Arch. OralBioi., 4 2 , 171 (1967). (21) C. W. Anderson and J. S. Kittelberger, manuscript in preparation. (22) A. Gee and V. R. Dietz, J. Amer. Chem. Soc., 7 7 , 2961 (1955). (23) H. Newesely, Monatsh. Chem., 98,379 (1967),
The Journalof Physical Chemistry, Voi. 78, No. 16, 1974
